Ethernet Video Series
Episode 01, Segment 04 of 05
Ethernet Tutorial (continued)
Remember our discussion about how an Ethernet segment is a lot like a rope,
stretched between two springs? When only two people place their hands on the
rope, we called that a “Point to Point network.” We said that several people
could place their hands on the rope and share it. Let’s call that a “Multi-Point
As you can imagine, it would be necessary for people participating in a Multi-Point network to be polite; to share the rope. Because if two people both started tugging on the rope at the same time, everybody would be confused. The messages would be garbled, and the use of the network would be wasted. To solve this problem, Ethernet designers decided again to borrow some techniques and etiquette from telegraphers.
Telegraphers came upon a natural way of sharing the wires stretched in
between their various cities by just using normal human behavior. When the wire
or circuit was in use, the telegrapher knew it because he could hear the
clicking and clacking of the telegraph sounder. In those cases, a polite
telegraph operator would simply conclude, “OK, I have to wait a little while
until the wire goes silent.”
Accordingly, the first rule of transmitting a message on Ethernet is to take turns, in this same orderly fashion. Stations are required to monitor the wire, and if it’s busy, they can’t transmit. They must wait for a little period of silence between messages before anyone can begin transmitting.
(Click above image for a larger version)
Still, even if everybody is well-behaved and waits for a polite period of silence on the wire before beginning to send a message, there’s always the possibility that two polite, different people might start transmitting at exactly the same time anyway, and they would still interfere with each other. To help solve this problem, the designers of Ethernet decreed that when anybody is transmitting a message on the wire, they must also monitor their own message, and make sure that they can receive it clearly- that it is not garbled. If two or more people do begin transmitting at the same time, as they monitor their own transmission, they will see that it is garbled and in that case, everybody is to immediately stop transmitting, and wait for a random short period of time before ever attempting again. This allows an orderly and efficient sharing of the network. In actual practice, it’s generally pretty easy for a station that wishes to transmit a message to simply wait for a silent period and then begin transmission. But once in a while, especially on Multi-Point networks with a lot of computers, two or more stations might try to transmit at the same time, in which case they all detect that there’s a problem, stop transmitting, and wait a random, short interval of time before any of them try it again. The conflicts are thereby resolved naturally.
Episode 01, Segment 5 of 5